Reply #3: Because everybody would have a slightly different answer and you couldn't analyze it

[AlGore-08.com]

Let's say you ask ten people: "Who did you vote for and why?"

Here are your answers:

1. Hillary Clinton. I like her best.
2. Barack Obama. His speeches speak to me.
3. John Edwards. He's the best one.
4. Hillary Clinton. I dunno, I didn't like any of them, really.
5. Hillary Clinton. She's the most experienced.
6. Barack Obama. He's totally hot.
7. Barack Obama. Because Hillary sucks.
8. Hillary Clinton. I thought the line about hurting her feelings was awesome.
9. Martin Van Buren. The alien overlords told me to.
10. Barack Obama. He's the only one who can win the general election.

With a sample as small as 10 people, you can give those kind of individual answers, but the sample isn't large enough to see a trend and make decisions based on those trends. You can't tell from this sample things like - - Are other folks voting for Obama just because they dislike Hillary Clinton? How many of them? Is it so many that Edwards' campaign might benefit by pointing out his differences with Hillary Clinton? Why do they dislike Hillary Clinton? What can the Clinton campaign do to win over the voters who are voting against her? Are ten percent of the voters really going to back the late President Van Buren?

If you asked everybody who voted in NH last night "Who did you vote for and why", you'd have about 400,000 different answers somebody'd have to read and interpret. It's harder to correctly interpret what "He's the best one" means than it is to tally the answers to set questions like "Which candidate is strongest on defense? Which candidate is strongest on education?" etc., etc.